Kubernetes

CL2–4

Connects by parsing your kubeconfig and confirming with cluster-info — it manages existing clusters, not EKS/AKS/GKE provisioning itself. Emits manifests, Helm charts, and Kustomize overlays.

Because native drift comparison strips those bookkeeping fields before comparing, a routine controller update to them never gets reported as drift, but any change to spec fields you actually declared does.

Kubernetes never emits Terraform — the code path that would build one deliberately throws.
Native drift detection strips resourceVersion, uid, managedFields, and status before comparing live objects to declared ones, so routine API churn doesn't look like drift.

Capability maturity

FinOps (cost sync)
Drift detection
Day-2 operations
Brownfield import
AI copilot

What NebulaStack actually generates for Kubernetes

The Kubernetes provider targets clusters you already operate: manifest and Helm generation, kubeconfig-based connection, drift via kubectl diff and typed client comparison, rollout wait/undo, informer watch, and applied-resource inventory. Provisioning managed clusters (EKS/AKS/GKE) is handled by the cloud provider's own Terraform module.

A minimal real resource

apiVersion: apps/v1
kind: Deployment
metadata:
  name: web
spec:
  replicas: 3

Under the hood

backend/src/services/k8s/**
backend/src/services/iac/k8s/**

Ready to try it?

Start for free, no credit card.

Start for free

Docker

AWS